* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Attitudes as Temporary Constructions
Survey
Document related concepts
Albert Bandura wikipedia , lookup
Impression formation wikipedia , lookup
Jon Krosnick wikipedia , lookup
Group dynamics wikipedia , lookup
Social perception wikipedia , lookup
Group polarization wikipedia , lookup
Interpersonal attraction wikipedia , lookup
False consensus effect wikipedia , lookup
Carolyn Sherif wikipedia , lookup
Introspection illusion wikipedia , lookup
Vested interest (communication theory) wikipedia , lookup
Social tuning wikipedia , lookup
Implicit attitude wikipedia , lookup
Elaboration likelihood model wikipedia , lookup
Self-perception theory wikipedia , lookup
Transcript
In me construction of social judmnents (pp. 37-65). Hillsdale, NJ: Erlbaum. (1992) L. L, Martin & A. Tesscr (Eds.), Attitudes as Temporary Constructions Timothy D. Wilson Sara D. Hodges University of Virginia For the past several years we have investigated the effects of introspection on attitude change. We have found that when people are asked to think about why they feel the way they do about something, they often change their minds about how they feel. People who analyze their reasons have been found to change their attitudes toward such diverse things as political candidates (Wilson, Kraft, & Dunn, 1989), dating partners (Wilson, Dunn, Bybee, Hyman, & Rotondo, 1984), art posters (Wilson, Lisle, & Schooler, 1990), food items (Wilson & Schooler, 1991), vacation pictures (Wilson e t al., 1984), and puzzles (Wilson & Dunn, 1986; for reviews, see Wilson, Dunn, Kraft, & Lisle, 1989 and Wilson, 1990). We have been struck by how easy it has been to get people to change their attitudes, in marked contrast to how unyielding people are in other kinds of social psychological studies. Our findings seem inconsistent with the vast literature on attitude change, which has found that you have to go to some length to change people's views. Consider research on persuasive communications. Only after listening to carefully crafted speeches that contain powerful arguments, or are delivered by attractive, expert sources, or both, do subjects change their attitudes (Eagly & Chaiken, 1984; Petty & Cacioppo, 1986). In our studies all we ask people to do is to think, privately and anonynlously , about why they feel the way they do-and they change their minds about how they feel. The ease with which people adopt new attitudes has caused us, along with some other recent theorists, to question the view that people hold stable attitudes. In this chapter we review evidence contrary to this tradition;^! viewpoint. 38 ' WILSON AND HODGES To preview, we suggest that people often have a large, conflicting "data base" relevant to their attitudes on any given topic, and the attitude they have at any given time depends on the subset of these data to which they attend. We review evidence showing that the data people use are influenced by both contextual factors and the kind of thought in which they engage. Finally, we argue that the attitude change resulting from these contextual and thought processes is sometimes detrimental, leading to nonoptimal preferences and decisions. Our characterization of attitudes as temporarily constructed judgments is quite consistent with the overall theme of this book concerning the construction of social judgment. It has become clear that many different kinds of judgments and impressions are influenced by the context in which they are formed. Most models of social construction, however, are concerned with situations in which people judge stimuli that are ambiguous or about which they know little (such as the priming literature, in which people form impressions of a person they know only from a brief, ambiguous paragraph describing his or her behavior). Attitudes have traditionally been viewed as more immune to changes in context, because they have been thought to be enduring predispositions that do not change from one situation to the next. In this chapter we extend models of social construction to the area of attitudes, arguing that attitudes can be profitably viewed as temporary constructions from whatever data are most accessible to people. THE TRADITIONAL VIEW OF ATTITUDE STABILITY Historically, att.itudes have been defined as evaluations that are stable over time (Cook & Flay, 1978). Allport (1935) noted that attitudes "often persist throughout life in the way in which they were fixed in childhood or in youth" (p. 814), whereas M. Sherif and Cantril(1947) suggested that "attitudes, once formed, are more or less enduring states of readiness" (p. 22). Petty and Cacioppo (1981) defined an attitude as an "enduring positive or negative feeling about some person, object, or issue" (p. 7). In Wilson, Lisle, and Kraft (19901, we referred to this as the file drawer analogy of attitudes: When people are asked how they feel about something, such as legalized abortion, their Uncle Harry, or anchovies on a pizza, presumably they consult a mental file containing their evaluation. They look for the file marked abortion, Uncle Harry, or anchovies, and report the evaluation it contains. The contents of these "files" may be changed by personal experiences, persuasive messages, and the like, but for the most part they are enduring evaluations that remain unchanged. We acknowledge that this file drawer analogy is simplistic, and underestimates the complexity of models of attitude structure. The exact nature of attitudes continues to generate considerable debate (see Pratkanis, Breckler, & Greenwald, 1989; Tesser & Shaffer, 1990; Zanna & Rempel, 1988 for reviews). 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 39 Despite the disagreements between different models of attitude structure, however, many of them assume that attitudes are stable, unchanging entities. That is, whatever the underlying structure of an attitude may be, the resulting evaluation is thought to remain relatively constant. People simply do not hate anchovies one day, only to spread them liberally on their pizza the next. A number of studies support this view, including many in the political domain that find considerable stability in people's political attitudes (e.g., Bennett, 1975; G . D. Bishop, Hamilton, & McConahay, 1980; Brown, 1970). For example, Marwell, Aiken, and Demerath (1987) found that the political attitudes of civil rights workers changed little over the course of 20 years. Other kinds of attitudes are also notorious for their resistance to change, such as prejudiced and racist opinions of minority groups. ATTITUDES AS CONSTRUCTIONS Despite this support for the idea that attitudes are stable constructs, there is another position gaining in popularity. According to this view there are no "true" attitudes. How people feel may depend on how they are asked and what they are thinking about at the time. Consider, for example, this statement by Abraham Tesser (1978): "an attitude at a particular point in time is the result of a constructive process. . . . there is not a single attitude toward an object but, rather, any number of attitudes depending on the number of schemas available for thinking about the objects" (pp. 297-298, emphasis in original). Other theorists have echoed and elaborated on Tesser's views (Schwartz, 1978; Schwarz & Strack, 1985; Tourangeau & Rasinski, 1988; Wyer & Hartwick, 1980; Zanna, 1990; Zanna & Rempel, 1988). Zanna and his colleagues, for example, argued that attitudes can be based on either affective, cognitive, or behavioral information, and vary depending on which of these three kinds of information is salient to people. Thus, it may not be very meaningful to ask how people feel about such issues as the welfare system or the death penalty. How they feel may depend on what kind of information they are currently using to construct an evaluation. Our position is similar to these recent conceptions of attitudes. We argue that: 1. People often construct their attitudes, rather than simply reporting the contents of a mental file. 2. When people construct their attitudes they have a large data base to draw from, including their behavior, their moods, and a multitude of (often contradictory) beliefs about the attitude object. 3. People rarely use the entire data base, instead constructing their attitude from a subset of these data. 40 WILSON AND HODGES 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 41 Sometimes people find tliemselves behaving in a way that is inconsistent with a prior attitude. As a result, they change their attitudes to match their behavior. We refer, of course, to research on dissonance and self-perception processes (Bern, 1972; Festinger, 1957). Subjects in an experiment by Cohen (1962), for example, were quite upset about how violently the police had behaved during a recent confrontation with students. After the experimenter asked the subjects to write an essay in support of the police for low external justification (i.e., little money), subjects moderated their views. This finding-one of the most replicated in social psychology-suggests that attitudes are, under some conditions, quite malleable. Despite the fact that people might have felt differently in the past, the datum they see as most relevant to how they feel now is often their own behavior, even when their behavior is caused by an external agent. & Rasinski, 1988; Turner & E. Martin, 1984). The way in which questions are asked can influence how people interpret which attitude the researcher wants to know, what information is relevant to that attitude, the standards of comparison people use, and how they choose to report their attitude. (See Strack & Martin, 1987, and Tourangeau & Rasinski, 1988 for excellent reviews of this literature.) Sometimes the way in which a question is asked influences the attitude people report, but not how they actually feel. Research on self-presentation has documented the powerful effects of situational variables on people's public reports of their feelings. A striking example of this was found in the 1989 gubernatorial race in Virginia, where many White voters were reluct;mt to tell pollsters tliat they favored Marshall Coleman, a White candidate, over Douglas Wilder, a Black candidate. The election day exit polls indicated that Wilder was ahead by 10 percentage points. He ended up winning by less than half a percentage point. White voters who were questioned by Black interviewers were especially reluctant to admit their support for Coleman (Shapiro, 1989). Such effects, while interesting in their own right, do not challenge the traditional view that attitudes are stable. Presumably, people's attitudes toward Coleman did not change; all that varied-from their responses to pollsters to their actions in the voting booth-was the attitude they chose to express. The important question for our purposes is whether the way in which survey questions are asked can influence how people actually feel. Though this is not a distinction that survey researchers have addressed very often, it is relatively clear that some kinds of context effects cause genuine attitude change. Tourangeau, Rasinski, Bradburn, and D'Andrade (1989a), for example, suggested that many people have contradictory beliefs about an issue such as welfare. Most people agree that the federal government has the responsibility to help needy people, but also that it is possible to succeed in America with determination and hard work. Tourangeau e t al. hypothesized that people's attitude toward welfare would depend on which of these beliefs was most accessible. Consistent with this prediction, people who first answered questions about the government's obligation to the needy expressed much more support for welfare than did those who first answered questions about individual determination (see also Judd, Drake, Downing, & Krosnick, 1991). Context Effects in Survey Research Accessibility a n d Impression Formation A common finding in survey research is that seemingly minor changes in wording or question order have large effects on people's responses to attitude questions (e.g., G . F. Bishop, 1987; Feldnian & Lynch, 1988; Fischhoff, Slovic, & Lirhtenstein, 1980; Hippler, Schwarz, & Sudman, 1987; Hogarth, 1982; Ottati, Riggle, Wyer, Scliwarz, & Kuklinski, 1989; Schuman & Kalton, 1985; Schuinnti & Pressc-r, 1981; T. Smith, 1987; Strack & I,. Martin, 1987; Tourangeau The Tourangeau et al. (1989a) study is reminiscent of another large literature, namely that on accessibility and impression formation. According to this literature, it is often difficult to form impressions of other people because tlieir behavior can be interpreted in different ways. Does the fact that John says little at the dinner party mean that he is shy? Depressed? Under the weather? Angry at the host? People's impressions of John will depend in part o n which of 4. The data people use are influenced by both the social context and the kind of introspection in which they engage. As a result, many attitudes are unstable, depending on the context and what people are thinking about. 5. The context and the content of people's introspections can lead to consequential changes in attitudes, changes that are sometimes nonoptimal. The careful reader will have noted a hedge in our argument, namely the statement that people "often" construct their attitudes. Although we argue that attitudes are often temporary constructions, we acknowledge that under some conditions people do have pre-packaged attitudes that do not have to be generated on the spot. We postpone a discussion of when this is likely to be the case. For now we examine the evidence for the position that attitudes are constructions, and as a result, can vary depending on the kind of data people use as building materials. CONTEXTUAL INFLUENCES ON T H E DATA P E O P L E USE T O CONSTRUCT T H E I R ATTITUDES Inferring Attitudes from One's Own Behavior 42 WILSON AND HODGES these categorizations are most accessible. Whether a particular concept is accessible depends on many things, such as how frequently a category has been used in the past (Higgins & King, 1981). Most relevant to our concern are times when a category is primed, temporarily and arbitrarily, by the context people are in, thereby causing people's impressions to vary from one context to the next. For example, if we had seen a television program about depression right before encountering John at the dinner party the construct of depression would be accessible, increasing the likelihood that we would think John is quiet because he is depressed (Higgins, Rholes, & Jones, 1977; Srull & Wyer, 1979). Models of accessibility have been concerned primarily with people's impressions of other people they have never met before (Higgins, 1989; Higgins & King, 1981; Wyer & SruU, 1980, 1989). It is important to note, however, that these models can be applied more generally to attitudes, be they about other people, physical objects, or social issues, as argued by Wyer and Srull(1989). Inferring Attitudes About Oneself from W h a t is Distinctive There is one class of attitudes that is thought by many to be especially immutable: people's attitudes about themselves. Throughout the history of psychology a prominent view has been that the self-concept is a unitary, stable, construct (e.g., Allport, 1937; Greenwald, 1980; Rogers, 1951; Swann, 1990). Others have suggested, however, that the self-concept is very much a social phenomenon, and varies according to the situation people are in (e.g., H. Markus & Kunda, 1986; Martindale, 1980; McGuire & Padawer-Singer, 1976; Mead, 1934; Sidis & Goodhart, 1904). James (1910), for example, suggested that a person has "as many different social selves as there are distinct groups of persons about whose opinion he cares" (p. 294). Recent experimental work has corroborated this claim. McGuire and Padawer-Singer found that people were likely to describe themselves in ways that were distinctive, as compared to their current social group. For example, when people who live in New York City are visiting Seattle, the fact that they are New Yorkers is distinctive, and thus is likely to be part of their self-concept. When back at home this fact is not at all distinctive, and thus will not be part of their self-concept. The fact that they live in Greenwich Village or Harlem or Staten Island sets them apart from most other New Yorkers, and thus is more likely to be part of the way they view themselves. In short, even our most important, basic attitude-our self-concept-depends on the data we use to construct it at any given point in time, and these data have been found to vary according to the context we are in. Inferring At,tit,ude's from One's Mood Another kind of datum people use to infer their attitudes is their current mood. Several studies have found that when people assess their overall satisfaction with tlieir lives, they are particularly influenced by their current mood states 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 43 (e.g., Schwarz, Strack, Kommer, & Wagner, 1987). The better their mood the greater their reported satisfaction with their lives, even though people know that moods vary from time to time. There are limits to these effects. If it is obvious to people that their mood is influenced by some arbitrary factor-such as the weather-they ignore their mood when reporting their overall life satisfaction (Schwarz & Clore, 1983). Summary Each of the literatures we have reviewed calls into question the assumption that attitudes are stable constructs. Each suggests that attitudes vary depending on the data people use to construct their attitude. Depending on the context, people have been shown to base their attitudes on their behavior, their mood, or a subset of their beliefs about the attitude object. What determines what kind of information people will use? Interestingly, each literature we have reviewed suggests that this depends on external, environmental events, such as someone inducing us to behave contrary to our beliefs, being in Seattle when describing oneself, or being asked about the government's obligation to the needy right before being asked about the welfare system. The implication is that if we were a hermit living in a cave our attitudes would remain stable. Because the social context would remain the same, so would our attitudes. E F F E C T S O F THOUGHT ON ATTITUDES We suggest that not even hermits have stable attitudes. Simply ruminating about one's attitudes and their causes can change the data people use to construct them. As we see in the following sections, two different kinds of thought have been found to lead to two different kinds of attitude change. Thinking About t h e Attitude Object Sometimes people ruminate about an attitude object, be it a social issue, a troublesome employee, or one's favorite baseball team. Research by Tesser and others has shown that such "mere" thought can lead to attitude change. Under some conditions thought leads to polarization, where people with a favorable attitude become even more favorable, and people with an unfavorable attitude become even more unfavorable. Polarization is likely to occur when people's beliefs are liigliev;iluatively consistent (i.e., all implying a pro or a con position). 'I'l~oii~l~t lights these beliefs, making the attitude more extreme (Cliaiken & Yiite~,1985; Judd & Lusk, 1984; Millar & Tesser, 1986b; Tesser, 1978). When people have several thoughts that are inconsistent with each other, thought leads to nioderation (Linville, 1982; Millar & Tesser, 1986b). The important point for our pur- 44 WILSON AND HODGES poses is that mere thought changes attitudes by changing the salience of the data people use to construct their attitude. Explaining the Reasons for One's Attitudes We have investigated a kind of thought that can also produce attitude change, but in a different way than merely think'ig about the attitude object. In our studies, people think about why they feel the way they do about something. As mentioned at the beginning of this chapter, we have found that introspecting about reasons often causes people to change their minds about how they feel. We now discuss the processes thought to be responsible for this effect, to show that it is consistent with the view that attitudes are constructions based on a subset of a large data base. People Are Often Unaware o f Exactly Why They Feel the Way They Do. The reasons people give for their attitudes are often incomplete or incorrect (Nisbett & Wilson, 1977; Wilson & Stone, 1985). Although some have found this conclusion to be too extreme (e.g., Ericsson & Simon, 1980; E. Smith & Miller, 1978), few would dispute the claim that people sometimes have difficulty knowing the exact determinants of their feelings. This is particularly true of our reactions to complex, multidimensional stimuli. For example, it is doubtful that we can explain exactly why we feel the way we do about other people. We have never heard anyone say, "I like Jane because I see her a lot," even though the effect of repeated exposure on people's attitudes is well established (Zajonc, 1968). Nor have we ever heard anyone say, "Because hostility is a category that is chronically accessible to me, I think John is a hostile person," even though the effects of chronic accessibility are also well documented (e.g., Bargh, Bond, Lombard!, & Tota, 1986). Wilson (1990) discussed several other influences on attitudes that people tend to overlook. The Reasons People Bring to Mind Can Implu a Somewhat Different Attitude. If people are unaware of exactly why they feel the way they do, what kinds of reasons do they bring to mind? One possibility is that people first access how they feel, and then bring to mind only those reasons that are consistent with their feelings. We suggest that this is most likely to occur when people have strong, accessible feelings. Many times, however, people's attitudes are not very accessible, and their search for reasons may be guided by a different search strategy. They may focus on those reasons that are accessible and plausible, regardless of whether they are consistent with their initial attitude. If a potential cause is plausible-that is, if it fits into people's causal theories about why they would like or dislike the attitude object-and if it is accessible in their memories, then they are likely to cite it as a reason. Attri- 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 45 butes are also likely to be cited as reasons if they are easy to verbalize or flattering to oneself. We should note that this view is compatible with some current models of person memory that suggest people use different kinds of memory searches when forming impressions (e.g., Carlston & Skowronski, 1986; Lingle & Ostrom, 1981). Carlston and Skowronski, for example, argued that people's impressions of others can be based on either their recall of prior impressions or recall of the person's specific behaviors. Similarly, we suggest that people use different strategies when thinking about why they feel the way they do. Sometimes their search for reasons will be guided by their past attitudes, particularly if these attitudes are highly accessible. Other times people are guided more by the plausibility and salience of different reasons, even if such reasons are not entirely consistent with their prior attitude. That is, if people's attitudes are not very accessible, they might focus on reasons that seem plausible but which imply a somewhat different attitude than they held before. Unaware That Their Reasons Are Biased, People Adopt the Attitude Implied bu Their Reasons. Because people are often unaware that the reasons they bring to mind are incomplete or biased, they assume that these reasons reflect how they feel. Earlier we argued that people have a large data base from which to construct their attitudes, but often use only a subset of these data. After thinking about why they feel the way they do, the data people appear to use are the reasons that come to mind. In our studies we examine people's reasons, and code them according to the attitude they imply. The correlation between this index and the attitude people report after analyzing reasons is usually very high. For example, it was .92 in a study by Wilson and Schooler (1991). There are several possible explanations for why we have found such high correlations between people's reasons and their subsequent attitudes. One not very interesting possibility is that people are fully aware of why they feel the way they do, and thus report reasons that are consistent with their prior feelings. The attitude they report after giving reasons is thus the same as they held before. Another possibility, as we have seen, is that people are not fully aware of the reasons for their attitude, but the reasons they call upon are still consistent with their feelings. For example, suppose people do not realize th;it they like a presidential candidate because of mere exposure (c.t;., they 11:ive seen the candidate on television many times). When asked why they feel the way they do, they say it is because he or she is trustworthy. Even if they are wrong about the role of the candidate's trustworthiness, focusing on this attribute implies the same, favorable attitude, and no attitude change results. T h e most interesting case occurs when people bring to mind one or more reasons that are evaluatively inconsistent with their prior attitude. As mentioned earlier, sometimes the reasons that are most plausible and accessible imply ;i new position. Suppose that when explaining their attitude people focus on the fact that, 46 WILSON AND HODGES now that they think about it, the candidate looks rather untrustworthy and excitable. We have found repeatedly that when this happens, people adopt the attitude implied by their reasons, resulting in attitude change. For example, Wilson, Kraft, and Dunn (1989) surveyed introductory psychology students' attitudes toward several possible Democratic and Republican presidential candidates in the spring of 1987. Several weeks later students came to the laboratory to participate in an ostensibly unrelated study. We asked half of them to write down why they felt the way they did about six of the candidates. As in most of our studies we told them that the purpose of this was to organize their thoughts, and that they would not be asked to hand in their reasons. To minimize self-presentational concerns even more, when the participants had completed the reasons questionnaire the experimenter said it would not be needed anymore, and deposited it in a trash can. People in a control condition completed a filler task of similar length. All subjects then rated their attitudes toward each of the candidates. The first question of interest is whether the reasons people brought to mind were consistent with their initial attitude. It turns out that they were, at least to some extent: The mean, within-subject correlation between the attitudes toward the candidates people expressed in their reasons and the attitudes they reported at the beginning of the semester was .71. Note, however, that this is not perfect consistency; in fact, the attitude expressed in people's reasons shared only half the variance of the attitudes reported earlier. In other words, there was some slippage between their prior attitude and their reasons. What about the attitudes toward the six candidates they reported right after listing reasons? One possibility is that people recognized that their reasons were biased or incomplete. If so, they might have disregarded the attitude implied by their reasons, reporting the same attitude as they had earlier. Our hypothesis was that people would fail to recognize that their reasons were biased or incomplete, and would use these reasons as the data to infer how they felt. Consistent with this hypothesis, the mean, within-subject correlation between the attitudes people expressed in their reasons and their subsequent attitudes was very high, M = .88. The difference between this mean and the one between attitudes expressed in reasons and Time 1 attitudes was nearly significant, /I = .06. These results, although consistent with our hypotheses, do not address directly whether attitude change occurred as a result of analyzing reasons. Before reporting these results, we need to digress a moment to discuss the kind of attitude change we would expect to occur. We have argued that people often bring to mind seasons that are (at least somewhat) inconsistent with their initial attitudes, and adopt the position implied by these reasons. But in what direction will this change be? Will tlie change be in a common direction, such that everyone becomes more positive or more negative, or will different people shift in different directions? Our position is that it depends on the kinds of reasons that 2 . ATTITUDES AS TEMPORARY CONSTRUCTIONS 47 come to mind. In some of our studies people bring to mind similar kinds of reasons for liking the attitude object, and thus shift in a common direction. For example, in a study by Wilson, Lisle, and Schooler (1990), people analyzed why they felt the way they did about different kinds of posters. Two of them were reproductions of paintings by Impressionist artists. As it happened, people found it easiest to bring to mind negative attributes of these paintings when analyzing reasons, and thus changed their attitude toward them in a negative direction. People found it easiest to bring to mind positive attributes of the other type of poster (humorous pictures of animals with captions), and thus changed their attitude toward these in a positive direction. Other times, however, analyzing reasons will not cause attitude change in a common direction. Instead, some people become more positive toward tlie attitude object, whereas others become more negative. One reason for this bidirectional attitude change is that different kinds of reasons are salient for different individuals. For example, when asked why they feel the way they do about a political candidate, people draw on different knowledge bases. The fact that is most salient to one person (e.g., that the candidate is anti-abortion) may be completely unknown to another. In addition, even if the same fact is available to everyone, such as the candidate's stance on abortion, it may be evaluated quite differently by different people, leading to attitude change in different directions. Consistent with these arguments, we found that people who analyzed reasons in the Wilson, Kraft, and Dunn study on political attitudes did change their attitudes, but not in a uniform direction. Some became more positive toward the candidates, others more negative. To assess this type of change we computed the absolute value of the difference between people's attitudes at Time 2 (after they had analyzed reasons) and at Time 1 (before they had analyzed reasons). As expected, the amount of absolute change was significantly higher for people who analyzed reasons than it was for people who did n0t.I '~arlierwe mentioned that the effects of merely thinking about an attitude object are different from the effects of analyzing the reasons for one's attitudes. It is possible, however, that these two kinds of thoughts have produced different effects only because different kinds of dependent measures have been used. We have found that analyzing reasons increases the absolute amount of change between a premeasure and a measure of attitudes taken right after the reasons analysis manipulation. This change is largest among people who are i~iiknowledgeahleabout tlic attitnde object (as discussed later in this chapter). Tesscr found that merely tliiiiking :ilwut tin- iittilnde object increases the percentage of people whose post-tliouglit attitudes are more extreme on tlie same end of the scale as their pre-thought attitude. This type of change is largest atnoiig people who are knowledgeable about the attitude object. It is possible that the key is not the different kinds of thought in wliicli people o i i g ; i ~ hut ~ , the way in which attitude change is measured. For exan~ple,if 'I't'sser's measure of polariz;ition were included in some of our reasons analysis studies, perhaps we would replicate his results. To test this possibility, we assessed polarization in several of our studies (e.g., Wilson, Kraft, & Dunn, 1989, Study 2). It turns out that analyzing reasons does not lead to attitude poliirization in either knowledgeable or unknowledgeable people. We also examined whether Tesser's mere thought manipulation causes changes on our absolute measure of attitude change that mirror tlie effects 48 WILSON AND HODGES 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 49 We have seen a considerable amount of evidence challenging the traditional assumption that attitudes are stable. Instead of reporting the contents of an attitude file, people often construct attitudes from the data that are most plausible and accessible. These data can be people's behavior, their moods, or particular beliefs about the attitude object that happen to be salient. The data people use :ire influenced by the social context and the kind of thought they engage in. Despite the evidence in support of these conclusions, they seem to belie common sense. Is it really the case that people are like chameleons, changing their attitudinal colors from one extreme to the other at the spur of the moment? It doesn't seem that way to us. Many of our attitudes have stayed the same for years. Both of us have had the same political party affiliation throughout our lives. One of us has always had a soft spot for old Dionne Warwick songs and a long-standing aversion to anything Barbra Streisand sings. The other has always loved going to the ballet. And neither of us has ever liked anchovies on our pizzas. It is possible, of course, that we are exceptions to the rule, rare people whose attitudes are especially fossilized. Although one of us admits that he has been accused of being an old curmudgeon, we trust that the reader shares our view that many attitudes seem very stable, contrary to the evidence that we have reviewed to this point. We believe there are several possible solutions to this paradox, to which we now turn. go. If the data that are accessible are negative, people will endorse a position at the negative end of their latitude of acceptance. According to this view they will never endorse a position in their latitude of rejection, which consists of positions they find unacceptable. This possibility has a certain intuitive appeal, because it explains the obvious fact that most people do not sway from one end of the attitude pole to the other according to the direction the contextual winds are blowing. If attitudes do vary only within a narrow range, however, the importance of the attitude-asconstructions model would be called into question. The bottom line would be that attitudes are relatively stable after all. Yes, you can make people endorse a slightly less liberal position than they endorsed before, but you cannot change a Democrat into a Republican simply by changing the context or altering how people think about political issues. We are not aware of any evidence that directly assesses the possibility that attitudes vary only within a narrow, prescribed latitude of acceptance. There is evidence, however, that contextual changes in attitudes are sometimes far from trivial. Schwarz e t al. (1987) found that people's moods had very large effects on their current life satisfaction, accounting for 24% of the variance. Tourangeau et al. (1989a) found that a minor change in the context of a survey changed people's support for increased spending for welfare from 62% to 44%, a difference of 18percentage points. Further, as we see later, the attitude change caused by the kind of thought people engage in can be large, and can cause people to make decisions that are nonoptimal. Attitudes Are Constructions, B u t Vary Only Within a Latitude of Acceptance Attitudes Vary i n Their Strength, a n d Weak Attitudes Are More Likely to Change with t h e Context Perhaps attitudes are constructed from the available data, but take on only a limited range of positions from one time to the next. For example, according to social judgment theory (M. Sherif & Hovland, 1961), people have a range of positions that they find acceptable, and the attitudes people construct might vary only within this range. Someone might be willing to endorse a variety of positions about George Bush, from the view that he has done a very poor job to the view that, on balance, he has succeeded in some areas and failed in others. The position endorsed at any given point in time will, as we have argued, be influenced by the data that are currently accessible. The latitude of acceptance, however, might provide boundaries on how far this construction process will Another possible solution is that some attitudes are stable, much like the file drawer analogy suggests. Others are labile, varying according to the context and what people are thinking about. To borrow an analogy used by Abelson (1986), some attitudes are like family heirlooms that we treasure and hold on to throughout our lives. Others are like a piece of clothing that we don according to the weather, our mood, and current fashion. This view is consistent with the idea we have just reviewed concerning people's latitude of acceptance, as long as we assume-as social judgment theory does-that the width of the latitude of acceptance is wider for some attitudes than others (e.g., C. Sherif, 1980). When people are very involved in an issue they have a narrow latitude of acceptance (and a correspondingly large latitude of rejection). Their altitude is c-leiirly defined, in that there are relatively few positions they are willing to endot st>. Perhaps these attitudes are the ones that remain stiible over time, and ;ire 1111mune to the effects of the context or what people are thinking about. When people are uninvolved in an issue they have a wide latitude of swept.-inw. T h e w HOW UNSTABLE A R E UNSTABLE ATTITUDES? AND ARE A L L ATTITUDES UNSTABLE? .... -. -. of analyziug reasons. 11 did not in two studies by Tessrr and I.conc (1977). (We thank Abraham Tesser for providing u s with the data from these studies.) Finally, some studies have included both types of thought manipulations and found that they have different effects (e.g., Tesser, Leone, & Clary, 1978). Thus, analyzing reasons and thinking about the attitude object appear to b e distinct kinds of tlioiiglit with (listinct kinds of effects on peopltx's attitudes. WILSON AND HODGES 50 attitudes may be more susceptible to change by the social context and the different kinds of thought people engage in. One diiculty with this hypothesis is that the concept of attitude strength is a thorny one, with several different meanings that appear unrelated to each other (Abelson, 1988; Krosnick & Abelson, in press; Raden, 1985; Wilson, Hedges, & Pollack, 1991). Nonetheless, there is evidence for the hypothesis that a strong attitude is a stable one (e.g., Converse, 1964; Fazio & Williams, 1986; Kendall, 1954; Krosnick, 1988; Petty & Cacioppo, 1986; Schuman & Presser, 1981; Taylor, 1983). Fazio and Williams, for example, found greater consistency between people's reports about whom they voted for in the 1984 presidential election and whom they said they supported 4 months earlier in people with accessible attitudes (as assessed with a reaction time procedure). Krosnick (1988) found that attitudes toward social issues that were important to people were less likely to change over the course of several months than were unimportant attitudes. Interestingly, very few studies have tested the corollary hypothesis that people with weak attitudes are more influenced by the context or the kind of thought they engage in.2 We have recently addressed this hypothesis in our research program on the effects of analyzing reasons on attitudes, at least indirectly. In the Wilson, Kraft, and Dunn (1989) study that looked at people's attitudes toward presidential candidates, we divided people into those who were knowledgeable about the candidates and those who were not. which was probably correlated with the strength of people's attitudes. Earlier we reported that the people who analyzed reasons were significantly more likely to change their attitudes. Actually, this was true only of people who were unknowledgeable about the candidates. As predicted, knowledgeable subjects were immune to the effects of analyzing reasons. The interaction between the reasons manipulation and people's amount of knowledge was significant. One reason that knowledgeable people were immune to the effects of analyzing reasons might be that they were more likely to recall their previous attitude, and were thus less likely to generate reasons that conflicted with it. Consistent with this view, the correlation between the attitudes subjects expressed in their reasons and their initial attitudes was higher for knowledgeable than unknowledgeable people (see Table 2.1). There are other possible interpretations of -- - ...- ' ~ nexception is a series of studies by Krosnick and Schuman (1988). They found in 27 surveys that people with intense, important, or certain attitudes were not less susceptible to such context effects as differences in question wording and response options. They suggested that the p:irlicular kinds of wording and response effects they examined did not cause genuine attitude change, t à § i instead influenced Irow people interpreted tlie questions ;ind how they chose to respond (i.e., self preseiitatioii;il concerns). There is no reason to assume lli;rt weak attitudes ;I!<* any more or less susceptible to such factors as self-presentational concerns or response biases. G . F. Bishop (1990) recently found evidence in support of the hypothesis we are considering here, namely that wr;il< ;ittit~~cles ;irr irrorc siisc<-11til)l(~ to ;ictual ;illil~i(l~(.II,III~(' in rcspnnsc lo c ~ ~ r t c xrffrcts. t 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 51 TABLE 2.1 Correlations Between the Attitudes Expressed 1 Reasons and Attitudes at Times 1 and 2 Low Knowledge High Knowledge Time 1 Attitude, Attitude Expressed in Reasons at Time 2 71 .8!i Time 2 Attitude, Attitude Expressed in Reasons at Time 2 88 Hi) Correlation Between: Ndr:'Froni Wilson, Kraft, and Dunn (1989, Study 2). the moderating effects of knowledge. Lusk and Judd (1988), for example, found that unknowledgeable people are more likely to have cognitions about the attitude object that are evaluatively inconsistent; that is, they are more likely to have a mixture of positive and negative beliefs. When unknowledgeable people analyze reasons, then, they are more likely to focus on at least some beliefs that conflict with their initial attitude, resulting in attitude change. Consistent with this interpretation, unknowledgeable people in the Wilson, Kraft, and Dunn (1989) studies were significantly more likely to generate reasons that were evaluatively inconsistent than were knowledgeable people. Attitudes Vary in Their Structure, and Some Kinds Are More Likely to Change with the Context The key variable moderating attitude stability may be not attitude strength, but attitude structure. In recent years there has been a renewed interest in the structure of attitudes (Pratkanis et al., 1989; Tesser & Shaffer, 1990). The questions of interest for our purposes are whether structural variables moderate how stable an attitude is, and how easily the attitude can be changed by changing the context or the kind of thought people engage in. As to the first question, there is evidence that more complex attitudes are more stable (at least when multidimensional measures of attitudes are used; see Schlegel & DiTecco, 1982). Similarly. Rosenberg (1960) found that attitudes with consistent affective and cognitive components are more stable than ones with inconsistent components. If consistent attitudes are the most stable, then we might expect that inconsistent ones would be most susceptible to the effects of the context and thought. There is a fair amount of evidence consistent with this proposition. F-or example, Cliaiken and Baldwin (1981) found that people wit11 inconsistent affective i g ~ attitudes toward tlir and cognitive components were most likely to ~ l i i ~ t their eiiviioiiinciit in response to a manipulation that nuidc either pro- or anti-ccology behaviors temporarily salient. People with consistent components did not change their nttitudes in response to this salience nianipt~lation.Siniil;irly, sevci ill in- 52 WILSON AND HODGES teresting studies by Millar and Tesser (1986a. 1989, this volume) found that people who have inconsistent affective and cognitive components are most susceptible to the effects of analyzing reasons. When people analyze reasons, they suggest, they focus on the cognitive component of their attitude; that is, rational thoughts about the attitude object (see also Wilson e t al., 1984). This can cause attitude change by turning what was an affectively based attitude into a cognitively based one. This attitude change, however, is only likely to occur to the extent that people's affect and cognitions are inconsistent with each other. If they are consistent, then focusing on one component versus the other will make little difference in how people feel. We have found some support for the hypothesis that analyzing reasons is most likely to change people's attitudes if their cognitions are evaluatively inconsistent (Wilson, Kraft, & Durn, 1989). We argue that when people analyze reasons they focus on a subset of their beliefs. The more inconsistent these beliefs are, the greater the likelihood that the subset people focus on will conflict with their previous attitude, leading to attitude change. (See our earlier discussion of the Wilson, Kraft, & Durn, 1989 study .) Finally, there is evidence that context effects in surveys are more likely to occur if people have inconsistent, multidimensional beliefs about the attitude object (Schuman & Presser, 1981). For example, if people have inconsistent beliefs about gun control-such as the idea that people have the right to bear arms, but also that lax gun control laws contribute to the high murder rate in this country-then their attitude toward gun control will be more influenced by questions on a survey that highlight one or the other of these beliefs. It is not entirely clear, however, how distinct structural variables such as the consistency or complexity of people's beliefs are from attitude strength (Abelson, 1988; Raden, 1985; Wilson et al., 1991). We can make a rough distinction between attitudes that are strong, complex, and made up of consistent feelings and cognitions versus those that are weak, not complex, and made up of inconsistent feelings and cognitions, and say that, with some exceptions, the former type of attitude is most likely to be stable and least likely to vary with the context. Very few studies, however, have examined whether the key moderating variable is strength, complexity, or consistency. An intriguing exception is a recent study by Tourangeau, Rasinski, Bradbum, and D'Andrade (1989b). They asked people how important several issues were to them, and how much they had mixed feelings and beliefs about these issues. They then observed how much an accessibility manipulation (i.e., questions in the survey that primed either positive or negative beliefs about the issues) influenced people's attitudes. Interestingly, the people who were most affected by this manipulation were those who had said that they had mixed feelings and that the issue was important. Thus, it may be that neither attitude strength nor structure by themselves moderate susceptibility to context effects. These two variables may interact in a more complex fashion. 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 53 Stable Attitudes Are Those with Stable Contexts Another possible solution to our paradox is that some attitudes are stable because the context in which we express them, or the way in which we think about them, typically does not change. Some attitudes, such as our feelings about the annual church picnic or a colleague we see only at work, are expressed primarily in one, unvarying context, thus we use the same, unvarying data to construct them. Further, people do not exert much time thinking about many attitude objects or trying to analyze why they feel the way they do. If so, these attitudes will remain constant not because they are strong or structurally consistent, but because the context in which they are expressed does not vary. Even if the context does vary, people might have a chronic way of constructing some attitudes, adding to their stability. That is, even if these attitudes vary according to the context people are in or what thoughts are accessible, people might "snap back" to a chronic way of construing the attitude object. This argument is similar to one made by Bargh et al. (1986; Bargh, Lombardi, & Higgins, 1988) about impression formation. These authors found that when people form impressions of others they have certain categories (e.g., shyness) that are chronically accessible, and that they view others in terms of these categories. The same may be true of some of people's attitudes more generally. If, over the years, we have found the annual church picnic to be dull and tedious, we are likely to construe it that way again this year (see Wilson, Lisle, Kraft, & Wetzel, 1989). We should stress that such attitudes are not immune to context effects. Our attitude toward the picnic is likely to be temporarily more positive if we are served a particularly delicious piece of fried chicken. The point is that, due to our chronic construction of such events, over time our negative attitude is likely to return. This possibility has some intriguing implications, which we discuss shortly. People Overestimate the Stability of Their Attitudes Finally, we should note that our examples about the constancy of our own attitudes about such things as Dionne Warwick and anchovies may be misleading, because there is a tendency for people to overestimate the extent to which their attitudes have remained constant (Bern & McConnell, 1970; Goethals & Reckman, 1973; G. Markus, 1986; Ross, 1989). Ross demonstrated that when people try to recall their past attitudes they assess how they currently feel and then consult their implicit theories about the stability of these feelings. Most people have the theory that attitudes do not change much over time (just as many social psychologists do), and therefore assume that how they feel now is how they have always felt. Because this theory is often incorrect, however, people overestimate how stable their attitudes have been. Thus, tlie fact that many of us can bring to mind examples of our own attitudes that have remained 54 WILSON AND HODGES the same over the years does not necessarily mean that these attitudes really have remained constant. Summary We began this section with a paradox. There is a considerable amount of evidence that attitudes are constructed from the available data. and thus vary from time to time depending on the data that are accessible. On the other hand, it simply doesn't seem like this is true, at least for many of our own attitudes. We offered several possible solutions to this conundrum, namely that attitudes may vary only within a latitude of acceptance, that only some types of attitudes are unstable (e.g., weak attitudes or those consisting of inconsistent beliefs), that people have a chronic way of constructing some attitudes, and that attitudes are less stable than people believe, due to memory biases. Although there is much more evidence supporting some of these solutions than others, we believe there is at least a grain of truth in all of them. DOES IT MATTER THAT ATTITUDES ARE SOMETIMES UNSTABLE? Some of the solutions we offered might appear to minimize the importance of the idea that attitudes are often unstable. For example, if attitudes vary only within a narrow range of acceptance, then the attitude change that results from a change in the context or in what people are thinking about will probably not be very consequential. We have seen, however, that such attitude change is often not trivial. We turn now to evidence that it can also cause changes in people's attitudes that are nonoptimal. We have already seen that when people think about why they feel the way they do, they often change their attitudes. We have recently posed the question of whether this change is beneficial, neutral, or harmful to people. Although it is unlikely to always be just one of these possibilities, we suggest that, at least at times, it can be harmful. There are two reasons for this. First, we assume that when people are left to their own devices, they often form satisfactory preferences and make good personal choices. People are certainly not perfect information processors, but they often manage to assign weights to the different attributes of the alternatives that produce a satisfactory choice (satisfactory to them). We assume that people often are not fully aware of how they are weighting the information, but often use schemes that work for them. The old adage that "I may not know why but I know what I like" probably has more than a grain of truth to it. If so, what happens when people introspect about why they feel the way they do? This kind of introspection, we suggest, can change an optimal weighting 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 55 scheme into a nonoptimal one. When people analyze reasons, we have argued that they focus on those attributes of the attitude object that are accessible in memory and seem like plausible causes of their evaluations. Those criteria that are accessible and plausible, however, might not have been weighted heavily before. Similarly, criteria that are not accessible or plausible-but were weighted heavily-will be overlooked. For example, when people evaluate a particular kind of food, they might be influenced by attributes of which they are unaware (e.g., an unknown ingredient) or that are difficult to verbalize (e.g., its texture or aroma). When asked to explain their reactions, they may assign less weight to these factors because they are difficult to put into words, and as a consequence change their preferences. Assuming that their original preference was fairly optimal, this change in weights might lead to a less optimal preference. The second reason that explaining one's attitudes might lead to nonoptimal choices concerns the extent to which the attitude change caused by analyzing reasons persists over time. We have argued that attitudes are often constructed from the available data, and thus vary according to which data are accessible. As argued earlier, however, people often have a chronic way of constructing their attitudes. These attitudes can be altered when people encounter new contexts or think about the attitudes in new ways, but otherwise remain fairly constant. If this argument is true, what happens when people analyze reasons? The new attitude that results might be particularly unstable because it is contrary to the way people chronically construct the data in that domain. For example, suppose someone is looking for something to hang over her mantie, and goes to a store that sells art posters. Suppose further that she is in a particularly introspective mood, and decides to analyze why she feels the way she does about each poster she examines. After doing so she decides which one she likes the best, purchases it, and takes it home. There is a good possibility, we suggest, that the act of analyzing reasons will change this person's mind about which poster to purchase. Trying to verbalize her reasons probably highlighted features of the posters that were not central to her initial evaluations, leading to a change in her preferences. Put differently, this person is probably not fully aware of how she usually (i.e., chronically) forms evaluations of works of art, leaving her open to the kinds of attitude change we have found so often in our studies. But what will happen over the next few weeks, as slie looks at the poster over her mantle? Our suspicion is that the features she focused on wlicn slie analyzed reasons would probably no longer be salient. Instead, she would revert back to her chronic means of evaluating works of art, possibly causing her to regret her choice. We recently tested whether this sequence of events can occur. Post-Choice Satisfaction with Consumer Choices We asked subjects to evaluate five posters of the type college students hang in their rooms, and then choose one to take home (Wilson, Lisle, & Scliooler, 1990). Two of the posters were reproductions of paintings by Impressionist 56 WILSON AND HODGES artists, and were very popular with our student population. The remaining three were humorous posters, such as a photograph of a cat perched on a rope with the caption, "Girnrne a Break." They were considerably less popular. In fact, in our control condition, where people were not asked to analyze reasons, 95% of the subjects chose one of the Impressionist paintings to take home. Half of the subjects analyzed why they felt the way they did before evaluating the posters. As expected, this introspection changed the preferences of many of these subjects. People who analyzed reasons tended to focus on positive attributes of the humorous posters and negative attributes of the Impressionist paintings. As a result, they were significantly more likely to choose one of the humorous posters (see Table 2.2). So far, these results are the same as those we have obtained in many other studies: Analyzing reasons changes people's attitudes. The interesting finding is what happened after people took their posters home. We telephoned subjects 3 weeks later and asked them several questions assessing how satisfied they were with the poster they had chosen. As predicted, and as seen in the bottom row of Table 2.2, subjects in the reasons condition reported a lower satisfaction with their choice of poster, possibly because their initial attitude-that the Impressionist paintings were preferable-had by then returned. According to our model, it should only have been people who convinced themselves they liked the unpopular humorous posters who later regretted their choice. As seen in Table 2.2, there was a tendency in this direction in that the difference in post-choice satisfaction between reasons and control subjects was greater anlong those who chose a humorous poster than those who chose an Impressionist poster. The interaction between condition and poster choice, however, was not significant. One reason for this might be that the control condition/humorous poster cell was represented by one person, making this value highly unreliable. We replicated the poster study with a new stimulus, felt-tip pens, and found results that were more consistent with our hypothesis. The people who were least satisfied with their choice of a pen were those in the reasons condition who convinced themselves that they liked one that was an unpopular color, and chose to take it home. When asked later how much they liked it, their original attitude seemed to have returned, making them relatively unhappy with their choice. The Wilson, Lisle, and Schooler (1990) studies have some intriguing implications about the effects of other kinds of context effects on post-choice satisfaction. When people have a chronic way of constructing a preference or attitude, then any factor that alters this construction might cause people to make decisions they later regret. That is, suppose people changed their attitudes due to any of the context effects we have reviewed, be they question order effects on surveys or the kinds of priming effects studied in the impression formation literature. If they made an important decision based on this new way of constructing their attitude, they might regret this decision later, when their chronic way of viewing the attitude object returns. 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS TABLE 2.2 Results of the Wilson, Lisle, and Schooler (1990) Poster Study Control Variable Choice (%) Post-Choice Satisfactiona Reasons Impressionist Humorous Impressionist Humorous 95 2.68 5 3.00 64 2.21 36 2.17 'Subjects were asked whether they still had the poster, whether they had hung it up, and whether they planned to take it home with them at the end of the semester. They received a zero if they said no to these questions and a one if they s;ii(l yes. The satisfaction index is tin- siiin of their three responses. Comparing People's Attitudes to Expert Opinion , People's reported satisfaction with a choice is open to a number of biases, such as a tendency to feel that whatever the choice, it was the best one to make (Brehm, 1956). To add weight to our claim that analyzing reasons can lead to nonoptimal decisions, we used a different criterion in two other studies: expert opinion about which alternatives are the best (Wilson & Schooler, 1991). In the first study we compared the preferences of subjects toward a food itemstrawberry jams-with the opinions of sensory panelists from Consumer Reports magazine. Subjects tasted five different brands. Half of the subjects analyzed why they liked or disliked each alternative, whereas half did not. All subjects then rated how much they liked each brand. We suggested earlier that, left to their own devices, people often make reasonably good decisions. Consistent with this hypothesis, the ratings made by control subjects corresponded fairly well with the experts' rating, resulting in a mean, within-subject correlation of .55. We have also suggested that trying to explain one's attitudes can influence the salience of certain attributes of the attitude object, causing people to change their evaluations. This prediction was also borne out, in that subjects who analyzed reasons ended up with significantly different preferences for tlie jams than did control subjects. Finally, consistent with our hypothesis that these prcferences would be in some sense "worse," the ratings made by subjects who analyzed reasons did not correspond with the experts very well, mean correlation = .11. The difference between the mean correlations in tlie control and reasons conditions was significant. The jam study examined people's preferences, without asking them to make an actual consumer decision. In a second study, Wilson and Sclioolcr (1991) examined a real-life choice of some import;ince to college students: the decision of which courses to take. A sample of introductory psychologv students, who had expressed an interest in taking more psychology classes, were seen at the beginning of the week when they registered for classes for t l i c next 58 WILSON AND HODGES semester. They were given a packet of information about each of the nine sophomore-level psychology classes being offered the next term. This packet contained such information as a description of the course content, the course evaluations of students who had previously taken the course, and whether a term paper was required. After reading through this information, subjects in the reasons condition were asked to describe why they might or might not take each course. Subjects in the control condition did not receive any special instructions about how to approach the course inf~rmation.~ After examining the information about the courses, subjects were given a surprise recall test in which they were asked to write down everything they could remember about the courses. We used their recall as a rough indication of how they had weighted the different attributes when forming their preferences, comparing it to the opinions of faculty members as to which criteria ought to be used when choosing a course. Control subjects were significantly more likely to recall information that faculty members rated as important (e.g., who was teaching the course) than information faculty members rated as unimportant (e.g., when the class met), suggesting that these subjects were weighting the information fairly optimally (at least to the extent that we can consider the faculty members to be experts in this domain). In contrast, subjects who analyzed reasons were no more likely to recall the important information than they were the unimportant information. We also examined (with subjects' permission) which courses they actually preregistered for. We assessed how good these choices were by comparing them to another kind of expert opinion: the course evaluations of students who had previously taken the courses. We assumed that it is advantageous for people to base their decisions of what courses to take on course evaluations, because these evaluations should predict how well they will enjoy the course. As discussed earlier, we assumed that, left to their own devices, people often make reasonably good choices. Control subjects were thus expected to base their choices, at least in part, on the course evaluations. We predicted that the reasons manipulation would change the criteria subjects used to make their choices, making them less likely to sign up for the highly rated courses. This prediction was confirmed, as seen in Table 2.3. Subjects in the control condition were significantly more likely to register for highly rated than for poorly rated courses. Subjects who analyzed reasons, however, showed only a slight preference for the highly rated courses. The Reasons Manipulation by Course Evaluation interaction was highly significant, p < .001. We also included a long-term measure of subjects' behavior: the courses they . -.. .---..-- - - - ' W e i~i,lulled;I Iliird condition ;IS well, wlu're sulijects were asked to introspect about how eiich and every piece of information about every course influenced tlieir preferences. For reasons of spiice we will omit a discussion of this condition here, except to note that w e expected this kind of introspection to alter people's preferences in nonoptiinal ways, albeit for different reasons. This prediction was confirmcil (':cc Wilson & Srhoolcr, 1991). 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 59 TABLE 2.3 Courses Preregistered for and Actually Taken, by Condition Vafiable Control Reasons :!7 .t 1:) 21 Preregistration Highly Rated Courses Poorly Rated Courses Actual Enrollment Highly Rated Courses Poorly Rated Courses 08 Note: Subjects were assigned a one if they registered for or actually took ;i course, and a zero if they did not register or take a course. Table adapted from Wilson and Schooler (1991), copyright 1991 by the American I'sychological Association. were enrolled in at the end of the following semester. Subjects had the opportunity to add and drop courses at the beginning of the semester; thus, even though the reasons manipulation influenced their initial decision of which courses to take, they could revise these decisions later. We did not make firm predictions about the outcome of this measure. On the one hand, we have argued that the attitude change caused by analyzing reasons is relatively temporary, and will not influence long-term behavior. Consistent with this view, Wilson et al. (1984, Study 3) found that analyzing reasons did not influence dating couples' decisions about whether to break up several months after the study was completed. On the other hand, if analyzing reasons changes subjects' decisions about the courses for which they register, they might experience some inertia, such that they remain in these courses, even if they change their minds at a later point. Further, Miar and Tesser (1986a, 1989) found that analyzing reasons highlights the cognitive component of attitudes, and that these cognitively based attitudes will determine behaviors that are more cognitively based than affectively based. Given that the decision of whether to take a college course has a large cognitive component (e.g., whether it will advance one's career goals), the attitude change that results from analyzing reasons might cause 1011gterm changes in behavior. Consistent with this latter possibility, by the end of the following semester subjects who had analyzed reasons were still less likely than control subjects to be enrolled in courses that were highly rated and more likely to lie enrolled in courses that were poorly rated (see Table 2.3). This effect had weakened over time, but the Reasons Manipulation by Course Ev;ilu;ilion interaction was still significant, p < .05.This is perhaps our strongest demonstration th;il ;in;ilyzing reasons can be costly, leading to nonoptiinal choices. We should address some possible ethical objections to the course selection study, given that it involved a consequential, real-life decision on the p;irt of 60 WILSON AND HODGES the participants. It might be argued that we should not have asked subjects to think about why they felt the way they did about each course, given our hypothesis that this kind of introspection would change the courses for which they preregistered, and possibly even change the courses they actually took the following semester. We thought about this issue at some length before conducting the study, and discussed it with several colleagues. In the end we decided that the potential knowledge gained-discovering some detrimental effects of introspection-outweighed the possible harmful effects on the participants. We should emphasize that we did not give the participants any misinformation about the courses-all of the information we gave them, including the course evaluations, was accurate. What we did was to ask some of them to reflect more than they might ordinarily do when forming their preferences. According to the predominant theories of decision making (e.g., Janis & Mann, 1977), asking people to be more reflective about their choices should have beneficial effects. There are probably many decision analysts, counselors, and academic advisors who urge people to make decisions more reflectively. Given that the effects of our manipulations were predicted to be relatively benign-altering the psychology courses for which subjects preregistered, and possibly altering the courses they took the following semester-we felt it was worth testing the wisdom of such advice. We did not, of course, make this decision alone. The study was approved by a Human Subjects Committee. SUMMARY AND CONCLUSIONS We began this chapter by posing the question of whether attitudes are stable entities or temporary constructions, fashioned from whatever data are accessible when people evaluate an attitude object. We reviewed a considerable amount of evidence supporting the attitudes-as-constructions hypothesis, namely research showing that people's attitudes vary according to the context and what they are thinking about. We then considered several moderator variables specifying when attitudes are likely to be stable, like files that we retrieve from our memories, and when they are likely to be recomputed from the accessible data. Although the evidence is mixed, there is some support for the hypothesis that strong attitudes with consistent components are likely to be stable, whereas weak attitudes with inconsistent components are most likely to vary with the context and the kind of thought people engage in. The bottom line seems to be that when we ask what people's attitudes are, we first need to consider the context in which they were asked, and what they were thinking about at the time. Not all attitudes will be affected by such variables-but enough will that it is worthwhile to ask these questions. Considering the context ;ind what people are thinking about is especially important because these variables can cause substantial chanfes in people's attitudes, and cause them to make nonoptimal derisions. 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 61 ACKNOWLEDGMENTS The writing of this chapter was supported by a grant from the National Institute of Mental Health (MH41841) to the first author and a National Science Foundation Graduate Fellowship to the second author. REFERENCES Abelson, R. P. (1986). Beliefs are like possessions. J11uma1fur thr 771(~ilrfnl Sncinl Itrhavii~ur. 16, 223-250. Abelson, R. P. (1988). Conviction. American Psycholo,~?ist,43, 267-275. Allport, G . W. (1935). Attitudes. In C. Murchison (Ed.), A handbook ofsocialfisyr~hnliify(pp. 798-844). Worcester, MA: Clark University Press. Allport, G. W. (1937). Pattern andgrowth in personality. New York: Holt, Rinrhart & Winston. Bargh, J. A., Bond, R. N., Lombardi, W. J., & Tota, M. E. (1986). The additive nature of chronic and temporary sources of construct accessibility. Jow-mil of Personality and Social P.q~cholofl, 50, 869-878. Bargh, J. A,, Lombardi, W. J., & Higgins, E. T. (1988). Automaticity of chronically accessible constructs in person x situation effects on person perception: It's just a matter of time. Journal of Personality and Social Psychology, 55, 599-605. Bern, D.J. (1972). Self-perceptiontheory. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 6, pp. 1-62). New York: Academic Press. Bern, D. J., & McConnell, H. K. (1970). Testing the self-perception explanation of dissonance phenomena: On the salience of premanipulation attitudes. Journal of Personality and Social 1'sychology, 14, 23-31. Bennett, W. L. (1975). Thepoliticalmindandthepoliticalenvironment. Lexington, MA: D . C. Heath. Bishop, G. D.,Hamilton, D. L., & McConahay, J. B. (1980). Attitudes and nonattitudes in the belief systems of mass publics. The Journal of Social Psychology, 110, 53-64. Bishop, G. F. (1987). Experiments with the middle response alternative in survey qucstiims. I'ublic Opinion Quarterly, 51, 220-232. Bishop, G. F. (1990). Issue involvement and response effects in public opinion surveys. Public Opinion Quarterly, 54, 209-218. Brehm, J. W. (1956). Post-decision changes in desirability of alternatives. Journal of Abnormal and Social Psychology, 52, 384-389. 0fiI11ioiiQiiinlrily, 3-1. Brown, S. R. (1970). Consistency and Hie persistence of idt-olofiy. 1'~/1lic 60-68. Carlston, D. E., & Skowronski, J. J. (1986). Trait memory and behavior memory: The effects of und Social alternative pathways on impression judgment response times. jmimiil r~tI'r~sniitilil~ Psychology, 50, 5-13. Chaiken, S., & Baldwin, M. W. (1981). Affective-cognitive consistency and tlie effect of salient behavioral information on the self-perceptionof attitudes. Journal of i'frsotiiilitj~unii Sm'ial l'syc h o l o ~ 41, , 1-12. i~idi~rcil :iltitinlr piilni i n Chaiken, S., & Yatcs, S. M. (1985). Attitude sclu-iti;ilirity an(l lli~~iiglil tion. Journal of I'ersonality and Social l'sychiilug~,49, 147014Hl. Cohen, A. R. (1962). An experiment on small rewards for discrepant compliance ;ind attitude change. (pp. 73 78). New In J. W. Brelim & A. R. Cohen (Rds.), Exfilnr(~tionsin rof;tiilii'r ~iis.~ni~niii'r York: Wilcy. 62 WILSON AND HODGES Converse, P. E. (1964). The nature of belief systems in mass publics. In D. E. Apter (Ed.), ldeology and discontent (pp. 206-261). London: The Free Press of Glencoe. Cook, T. D., & Flay, B. R. (1978). The persistence of experimentally induced attitude change. In L. Berkowitz (Ed.), Advances in experimental socialpsychology(Vol. 11, pp. 1-57). New York: Academic Press. Eagly, A. H., & Chaiken, S. (1984). Cognitive theories of persuasion. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 17, pp. 267-359). Orlando, FL: Academic Press. Ericsson, K. A., & Simon, H. A. (1980). Verbal reports as data. Psychological Review, 87,215-251. Fazio, R. H.. & Williams, C. J. (1986). Attitude accessibility as a moderator of the attitude-behavior relation: An investigation of the 1984 presidential election. Journal of Personality and Social Psychology, 51, 505-514. lÂ¥'eldman J. M., & Lynch, J. G., Jr. (1988). Self-generated validity and other effects of measureinent on belief, attitude, intention, and behavior. Journal of Applied Psychology, 73, 421-435. 1"estiiiger. L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University Press. Fischhoff, B., Slovic, P., & Lichtenstein, S. (1980). Knowing what you want: Measuring labile values. In T. S. Wallsten (Ed.), Cognitiveprocesses in choice and decision behavior (pp. 117-141). Hillsdale, NJ: Lawrence Erlbaum Associates. (iocthals, G . It., & Recknian, R. F. (1973). The perception of consistency in attitudes. Journal of Exf~t-rinirnt111 Social Psychology, 9, 491 -501. (ireenwald, A. G . (1980). The totalitarian ego: Fabrication and revision of personal history. American I'sycholofiist, .?5,603-618. lliggins, E. T. (1989). Knowledge accessibility and activation: Subjectivity and suffering from unconscious sources. In J. S. IJleman & J. A. Hargh (Eds.), Unintended thought (pp. 75-123). New York: Guilford I'ress. IIiggi~is,E. T . & King, G. (1981). Accessibility of social constructs: Information-processing consequences of individual and contextual variability. In N. Cantor & J. F. Kihlstrom (Eds.), Personality, cognition, and social interaction (pp. 69-121). Hillsdale, NJ: Lawrence Erlbaum Associates. Higgins, E. T., Rholes, W. S., &Jones, C. R. (1977). Category accessibility and impression formation. Journal of Experimental Social Psychology, 13, 141-154. Hippler, H. J., Schwarz, N., & Sudman, S. (Eds.). (1987). Social information processing and survey methodology. New York: Springer-Verlag. Hogarth, R. M. (1982). Question framingand response contingency. San Francisco: Jossey-Bass. James, W. (1910). Psychology: The briefer course. New York: Holt. Jnnis, I. L., & Mann, I-. (1977). Decision making: A psychological analysis of conflict, choice, and commitment. New York: The Free Press. Jiidd, C. M., Drake, It. A., Downing, J. W., & Krosnick, J. A. (1991). Some dynamic properties of attitude structures: Context induced response facilitation and polarization.Journal ofPersonality and Serial Psychology. 60, 193-202. Judd, C. M., & Lusk, C. M. (1984). Knowledge structures and evaluative judgments: Effects of structural variables on judgment extremity. Journal of Personality and Social Psycholofg, 46, 1193 -1207. of respiinsr. (!I('IK'o(~, 11,: '1'lita Free KcII(I:III,1'. ( I !)fi'l). ( 'i~njliclunil niimd: Fiiclm's /~//(Â¥cli .'Â¥lal~ilit I'rcss. Krosnick, J . A . (1988). Attitude importance and attitude change. Journal of Experimental Social Psychulofff, 24, 240-255. Kmsnick, J. A.. & Abelson, R. P. (in press). The case for measuring attitude strength in surveys. In J. 'kiiiur (Ed.), Questions about survey questions. New York: Russell Sage. Kiosnick, J . A., & Schuman, 11. (1988). Attitude intensity, importance, and certainty and susceptibility to response effects. Journal of Personality and Social Psychology, 54. 940-952. 1-inglc,J . I!., & Ostroin, T. M. (1981). I'rinciples of memory and cognition in attitude formation. In R. E. Petty, T. M. Ostrom, & T. C. Brock (Eds.), Cognitive responses in persuasion (pp. 399-459). Hillsdalc. NJ: Lawrence Er1h:ium Associates. 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 63 L'iville, P. W. (1982). The complexity-extremity effect and age-based stereotyping.Journal nfPersonality and Social Psychology, 42, 193-211. Lusk, C. M., & Judd, C. M. (1988). Political expertise and the structural mediators of candidate evaluations. Jmimal of Experimental Social Psychology. 24, 105-126. Markus, G. (1986). Stability and change in political attitudes: Observed, recalled, and "explained." Political Behavior, 8, 21-44. Markus, H., & Kunda, Z. (1986). Stability and malleability of the self-concept.Journal of Prrsonality and Social Psychology. 51, 858-866. Martindale, C. (1980). Subselves: The internal representation of situational and personal dispositions. In L. Wheeler (Ed.), Review of personality and social ftsyfholosy (Vol. 1 , pp. 193-218). Beverly Hills, CA: Sage. Marwell, G., Aiken, M., & Deinerath, N. J. (1987). The persistence of political :ittitiuk~samoii~ 1960s civil rights activists. Public Opinion Quarterly, 51, 359-375. McGuire, W. J., & Padawer-Singer, A. (1976). Trait salience in the spont;ineous self-concept.Joiinuil of Personality and Social Psychology, 33, 743-754. Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago Press. Millar, M. G., & Tesser, A. (1986a). Effects of affective and cogiulive fw:us 011tlic :illili~~lr-l~cl~:ivio relationship. Journal of Personality and Social Psychology, 51, 270-276. Miillar, M. G., & Tesser, A. (1986b). Thought-induced attitude change: The effects of sclicinn stnicture and commitment. Journal of Personality and Social Psycholofy, 51, 259-269. briar, M. G., & Tesser, A. (1989). The effects of affective-cognitive consistency and thought o n the attitude-behavior relation. Journal of Experimental Social Psychology, 25. 189-202. Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on inentnl processes. Psychological Review, 84, 231-259. Ottati, V. C., Riggle, E. J., Wyer, R. S., Schwarz, N.,& Kuklinski, J. (1989). Cognitive and affective bases of opinion survey responses. Journal of Personality andSocial Psychology, 57,404-4 15. Petty, R. E., & Cacioppo, J. T. (1981). Attitudes and persuasion: Classic and rontcntpor(i~@preaches. Dubuque, IA: Brown. Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes to attitude chmge. New York: Springer-Verlag. Pratkanis, A. R., Breckler, S. J., & Greenwald, A. G. (Eds.). (1989). Attitude structure aniifiinction. Hillsdale, NJ: Lawrence Erlbaum Associates. Raden, D. (1985). Strength-related attitude dimensions. Social l'sycholngy Quarterly. 48. 312 330. Rogers, C. R. (1951). Client centered therapy. New York: Houghton-Mifin. Rosenberg, M. J. (1960). A structural theory of attitude dynamics. Public Opinion Qu/irtrrlj, 24, 319-341. Ross, M. (1989). Relation of implicit theories to tlie construction of pri'soii;il histories. 1'syrlmlof:icat Review, 96, 341-357. Schlegel, R. P., & DiTecco, D. (1982). Attitudinal structures and the attitude-behavior relation. In M. P. Zanna, E. T,Higgins, & C. P. Herman (Eds.), Consistency in social behavior: 7'hr Ontario Symposium (Vol. 2, pp. 17-49). Hillsdalc, NJ: Lawreii(-cErlliiu~~ii Associ.ites. Schuman, 11.. & Kalton, G. (1985). Survey methods. In (;. I,indzi-y A K. Aronsoii (1Cd.s.). Ilinnl book of social psychology (3rd ed., Vol. 1, pp. 635fi97). New York: R:indoni House. Schuman, H., & Presser, S. (1981). Questions and i i n s n ~in~tiltitmir sunvys. New Yolk: A G demic Press. Schwartz, S. H. (1978). Temporal instability as a moderator of the attitude-behavior relationship. Journal of Personality and Social Psychology, 36, 715-724. Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judpncnt of well-being: Inforinative and directive functions of affective states. Joumiil of ['ersonnlily (inif.%)riiil I'sychidofy; 45, 513-523. Schwarz, N., & Strack, F. (1985). Cognitive and affective processes in judgments of well-being: A preliminary model. In H. Brandstatter & E. Kichler (lids.), Economicpsyrholo~(pp. 439-447). Linz, Austria: Taubler. 64 WILSON AND HODGES Schwarz, N., Strack, F., Kommer, D., & Wagner, D. (1987). Soccer, rooms, and the quality of your life: Mood effects on judgments of satisfaction with life in general and with specific domains. European Journal of Social Psychology, 17, 69-79. Shapiro, W. (1989, November 20). Breakthrough in Virginia. Time, pp. 54-57. Sherif, C. W. (1980). Social values, attitudes, and involvement of the self. Nebraska Symposium on Motivation, 27, 1-64. Slierif, M., & Cantril, H. (1947). Thepsychology of ego-involvements: Social attitudes and identifications. New York: Wiley. Slierif, M., & Hovland, C. I. (1961). Social judgment: Assimilation and contrast effects in communicalion and attitude change. New Haven, CT: Yale University Press. Sidis, B., & Goodhart, S. P. (1904). Multiplepersonality: An experimental investigation into the nature of human individuality. Englewood Cliffs, NJ: Prentice-Hall. Smith, E. R., & Miller, F. 1). (1978). Limits on perception of cognitive processes: A reply to Nisbett and Wilson. Psychological Review, 85, 355-362. Smith, T. W. (1987). That which we call welfare by any other name would smell sweeter: hi analysis of the impact of question wording on response patterns. Public Opinion Quarterly, 51.75-83. Smll, T . K., & Wyer, R. S. (1979). The role of category accessibility in the interpretation of information about persons: Some determinants and implications. Journal of Personality and Social Psychology, 37, 1660-1672. Strack, F., & Martin, L. L. (1987). Thinking, judging, and communicating: A process account of context effects in attitude surveys. In H. J. Hippler, N. Schwarz, & S. Sudman (Eds.), Social information processing and survey methodology (pp. 123-148). New York: Springer-Verlag. Swann, W. B., Jr. (1990). To be adored or to be known? The interplay of self-enhancement and self-verification. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 408-448). New York: Guilford Press. Taylor, M. C. (1983). The black-and-white model of attitude stability: A latent class examination of opinion and nonopinion in the American public. American Journal o/Sociology, 89,373-401. 'I'esser, A. (1978). Self-generated attitude change. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 11, pp. 289-338). New York: Academic Press. Tesser, A., & Leone, C. (1977). Cognitive scheinas and thought as determinants of attitude change. Journal of Experimental Social Psychology, 13, 340-356. Tesser, A,, Leone, C., & Clary, G. (1978). Affect control: Process constraints versus catharsis. Cognitit~eTherapy and Research, 2, 265-274. 'I'esser, A., & Sliaffer, U. (1990). Attitudes and attitude change. Annual Review of Psychology, 41, 479-.523. Toiirangeau, It., 81 Rasinski, K. A. (1988). Cognitive processes unclerlying context effects in attitude measurement. Psychological Bulletin, 103, 299-314. Toiirangeau, R., Rasinski, K. A., Bradbum, N., & D'Andrade, R. (1989a). Belief accessibility and context effects in attitude measurement. Journal o/Experimental Social Psychology, 25,401421. Toiirangeaii, R., Rasinski, K. A., Bradbum, N., & D'Andrade, R. (198%). Carryover effects in attitude surveys. J'uhlic Opinion Quarterly, 53, 495-524. Turner, C. I:., & Martin, E. (Kds.). (1984). Sunwying subjective phennmnui (Vol. 1). New York: Russell Sige Foundation. Wilson, T. D. (1990). Self-persuasionvia self-reflection. In J. Olson & M. P. Zanna (Eds.), Selfinference processes: The Ontario Symposium (Vol. 6, pp. 43-67). Hillsdale, NJ: Lawrence Erlbiiiim Assoriates. Wilson, T. I)., & Dunn, I). S . (1986). Effects o f introspection o n attitude-behavior consistency: Analyzinfi reasons versus focusing on feelinrs. Journal of Exp(~timrnta1Social I'syrholney, 22, 249- 263. Wilson, T. D., Dunn, D. S., Bybee, J. A., Hyman, I). B., & Rotondo, J. A. (1984). Effects of analyzing reasons on attitude-beliaviorconsistency. Jownal of Personality and Social Psycholo47. 5. I(;. *, 2. ATTITUDES AS TEMPORARY CONSTRUCTIONS 65 Wilson, T. D.,Dunn, D. S., Kraft, D., & Lisle, D. J. (1989). Introspection, attitude change, and attitude-behaviorconsistency: The disruptive effects of explaining why we feel the way we do. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19, pp. 123-205). Orlando, FL: Academic Press. Wilson, T. D., Hodges, S. D., & Pollack, S. E. (1991). Effects of explaining attitudes on survey responses: The moderating effects of attitude accessibility. Unpublished manuscript. University of Virginia, Charlottesville. Wilson, T. D., Kraft, D., & Dunn, D. S. (1989). The disruptive effects of explaining attitudes: The moderating effect of knowledge about the attilmlo object. Joiirnnl 1 ~ f ~ i ~ ~ r i ~ 1 1Soriiil rfitnl Psychology, 25, 379400. Wilson, T. D., Lisle, D.J., & Kraft, D. (1990). Effects of self-reflectiim o n :ittitii(lcs;IIKI cii~~si~nir decisions. Advances in Consumer Research, 17, 79-85. Wilson, T. D., Lisle, D. J., Kraft, D., & Wetzel, C. G. (1989). Preferences as expectation-driven inferences: Effects of affective expectations on affective experience. Journal oj l'ersonality and Social Psychology, 56, 519-530. Wilson, T. D., Lisle, D. J., & Schooler,J. (1990). Some undesirable effectsof srlf-r~~lrriion. Unpublished manuscript, University of Virginia, CharlottesviUe. Wilson, T. D., & Schooler, J. (1991). Thinking too much: Introspection can reduce the quality of preferences and decision. Journal of Personality and Social Psychology, 60, 181-192. Wilson, T. D., & Stone, J. I. (1985). More on telling more than we can know. In P. Shaver (Ed.), Review of personality and social psychology (Vol. 6. pp. 167-183). Beverly Hills, CA: Sage. Wyer, R. S., & Hartwick, J. (1980). The role of information retrieval and conditional inference proresses in belief formation and change. In L. Berkowitz (Ed.), Advances in rxperimrntal social psvchology (Vol. 13, pp. 241-284). Orlando, FL: Academic Press. Wyer, R. S., & Srull, T. K. (1980). The processing of social stimulus information: A conceptual integration. In R. Hastie, T. Ostrom, E. Ebbesen, R. Wyer, D. Hamilton, & D. Carlston (Eds.), Person memory: The cognitive basis of social perception (pp. 227-300). Ilillsdale, NJ: Lawrence Erlbaum Associates. Wyer, R. S., & Srull, T. K. (1989). Memory andcognition in its socialcontext. IIillsAilc. N J : Lawrence Erlbaum Associates. Zajonc, R. B. (1968). Attitud'ial effects of mere exposure. Jottnr(;/ of f'rrsonnlfi unit Sorial J'sychology Monograph Supplement, 9, 1-28. Zanna, M. P. (1990). Attitude functions: Is it related to :ittituric structure? A(11~i111rrs in (.'rnsiinirr Research, 17, 98-100. Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old concept. In 1). Bar-Tal & A. W. Kruglanski (Eds.), The socialpsychology ofknowledge (pp. 315-334). Cainbridge, Engl;ind: Cambridge University Press.